The driver-assistance system has to estimate the driver’s head pose to prevent road hazards. The aim is to focus on prevention of road accidents due to driver fatigue or distraction under challenging lighting conditions. The proposed system defines the driver’s direction of attention and head-nodding detection. Head pose usually represents the focus of attention. Interaction of driver-assistance system with human requires accurate perception and tracking of their body parts to infer non-verbal signals of intention and attention. Head pose estimation is also used for analyzing complex meaningful gestures such as head nodding or pointing gestures. The pose estimation approach takes the input image. From the input image, face and features are detected using modified Viola-Jones algorithm. The combination of geometric and tracking method (hybrid technique) is used to estimate the head pose of the given input image. Training is not required for this approach. Finally, the experiment results on publicly available pose datasets or real-time images in driving scenario show that proposed algorithm achieves high performance and estimate the head pose with speed and higher robustness which is invariant to illumination conditions.
Estimating Driver Attentiveness Through Head Pose Using Hybrid Geometric-Based Method
Smart Innovation, Systems and Technologies
2022-04-19
8 pages
Article/Chapter (Book)
Electronic Resource
English